Perl was built initially as a sed/awk killer but got distracted into trying to take over the world. The interpreter for a language with 100x the number of features will always be slower. Also there's a very clear boundary for when I should use awk by itself, as part of a pipeline, or switch to a better tool. I feel like Perl has the potential to suck me imperceptibly into a huge mess where I spend 80% of my time refactoring everything.